AnchorShell Relay

Save on AI costs.
Keep every agent
on budget.

Set request, token, and spend limits by user, agent, model, or provider—then see every request and what it cost.

Get startedSee how it works

Shared AI budget

Example policy: a coding agent has reached its five-dollar daily limit and is blocked, while a support agent remains within its twenty-dollar limit and continues running.

The AI gateway that just works.

Opinionated defaults for queues, limits, routing, fallbacks, team controls, and observability—ready out of the box.

  • Request, token & spend limits
  • Per-user & per-model budgetsManaged
  • Queueing & pacing
  • Rate limiting & cooldowns
  • Provider fallbacks
  • Local/free-first routing
  • Provider & routing permissionsManaged
  • Usage & cost analytics
  • Full request history
  • Optional payload capture
  • Team roles & permissionsManaged
  • OpenAI-compatible endpoint
Explore all Relay features

Keep your SDK.
Change the base URL.

Point your existing OpenAI-compatible client at AnchorShell Relay. Your application keeps making model calls the same way.

One line deploy

- baseURL: "https://api.openai.com/v1"
+ baseURL: "https://api.anchorshell.com/v1"
Limits, queues, routing, fallbacks, and request history now sit behind the endpoint.

Relay sits between your AI application and the providers it calls.

Relay sits between an AI application and the providers it calls.

  1. 1
    Your app or agent

    Coding agent · OpenClaw · Chat interface · AI harness · Internal workflow

  2. 2
    AnchorShell Relay

    Limits · Queue · Route · Log

  3. 3
    Provider or aggregator

    Cloud provider · Free-tier endpoint · Local server · Model aggregator

  4. 4
    LLM

    Local or hosted model

Control exactly who can use what.

Set request, token, or spend limits for everyone, one user or agent, one model, or one provider. Choose a window from one second to one month.

Give expensive models tighter limits. Stop one agent without stopping everyone else.

Set a limit

1 What to controlRequestsTokensSpend
2 How oftenSecondMinuteHourDayMonth
3 Who or whatEveryoneUser or agentModelProvider

Policy summary

TargetCoding agent
ModelGPT-4.1
Limit$5 per day
$5.00 usedNext request blocked

Estimated token check

Remaining daily allowance18,000 tokens
Estimated request4,200 tokens

within limit

Remaining2,000 tokens
Estimated request4,200 tokens

blocked before dispatch

Block new requests before their estimate exceeds the remaining allowance.

Give every teammate a budget—and the right level of access.

AnchorShell Relay — Managed lets you invite users, control what they can see, and set request, token, and spend limits by user, model, or provider.

  • Attribute spend to a person or agent.
  • Stop one user without stopping the organization.
  • Separate own usage from organization visibility.
  • Manage Relay access without one unrestricted key.

Team

AnchorShell Relay — Managed
Managed

Team Members

3 active

User controls

Dave · Developer
Active
Requests
100 / day
Tokens
100,000 / day
Spend
$5.00 / day
Model-specific limits
3 configured
Usage visibility
Own usage
Role presets provide a starting point. Managed administrators can assign custom Relay permissions and user-specific request, token, spend, model, or provider limits.

See how every request moves through Relay.

Watch requests wait, move through an available model path, and complete with their owner, tokens, and cost attached.

Rank local and free paths first, let cooling paths wait, and keep an eligible cloud provider ready when needed.

Realtime request flow

One request moving through Relay
Flow active

Request rq_214 waits in the queue, skips a cooling free endpoint, moves through the available local model, and completes with Maya as owner, 1,284 tokens, and four cents in cost.

Queue

rq_214Maya · waiting
rq_215Research · waiting
rq_216Coding agent · waiting

Available models

Local modellocal/worker
available
Free endpointfree/general
cooling
Cloud providercloud/standard
available

Cooling path skipped

Completed requests

rq_214completed
Owner
Maya
Tokens
1,284
Cost
$0.04

Local model · 1.3s total

Know where every token and dollar went.

Start with the usage trend, filter down to the user or model, then open the exact request that caused it.

Break usage down by user, model, provider, and time range so you can see where capacity and budget are going.

Usage over time

Requests, tokens, and spend across Relay
Requests usage over the selected dayA line chart with one main peak, a smaller later rise, and the latest data point selected.9006003000Latest86 requests9 AM12 PM3 PM6 PMNow

Inspect every model API call.

Relay records the owner, model, route, status, tokens, cost, and timing behind each routed request.

Turn full payload capture on for deep debugging, or keep only operational metadata for a lighter footprint. Body capture is a global Relay setting and is off by default.

Request history

Operational metadata for routed model calls
Metadata recorded
Request history with rq_7F2B selected for inspection
RequestOwnerRouteResultTokensTimingCost
rq_7F2ACoding agentlocal/workerBlocked—42 ms—
rq_7F2BMayacloud/gpt-4.1Completed2,3181.3 s$0.04
rq_7F2CResearchfree/generalCompleted860920 ms$0.00

Request details

rq_7F2B
Completed
Request ID
rq_7F2B
Owner
Maya
Model
gpt-4.1
Provider
Cloud provider
Status
Completed
Input tokens
1,906
Output tokens
412
Cost
$0.04
Queue time
184 ms
Duration
1.3 s

Store request and response bodies

Global Relay setting · off by default
Metadata only

New request and response bodies are not retained while capture is off.

Fewer failed requests when a provider has a bad day.

Keep backup providers configured. When one path is throttled, unavailable, or fails, Relay can move eligible requests to the next option.

Fewer outages. Less fallback code to maintain.

Provider fallback

Configured model path
Recovered
Requested routegeneral-purpose
first available path
  1. Primary modelcloud/primary
    Provider statusUnhealthy
    skipped
  2. Regional backupcloud/secondary
    Provider statusAvailable
    selected
  3. Local modellocal/worker
    Provider statusAvailable
    ready
  4. Reserve modelcloud/reserve
    Provider statusAvailable
    ready
Relay skips the unhealthy path and continues through the next available configured model.

Put the limits in place before the next agent runs.

Use AnchorShell Relay — Managed, or run AnchorShell Relay — Self-Hosted on your infrastructure. Keep one OpenAI-compatible application endpoint either way.

Get startedSelf-host Relay